1. Field of the Invention
The present invention relates to a grid moving method for minimizing image information of an object image and an apparatus therefor, and a compaction/motion estimation method using the grid moving method and an apparatus therefor, and particularly to an improved grid moving method for an object image and an apparatus therefor and a compaction/motion estimation method using the grid moving method and an apparatus therefor, which are capable of forming a grid with respect to an image of a predetermined object having shape information and, in a region of an image, dividing the region of the image into a plurality of unit regions, moving the formed grid and detecting a position at which the amount of information is reduced when performing a compaction or estimating a motion of an object. In addition, the present invention is basically directed to reforming the grid to a position at which the amount of information is reduced by moving the grid, and is directed to separating and coding each unit region in which an image of the object exist from the reformed grid and using the grid movement of the image of an object of which the motion is estimated.
2. Description of the Related Art
Conventionally, since an image of an object having a predetermined shape contains a great amount of image data, when storing the image data in a recording/writing medium, a large space is necessary for storing the data. In addition, transmitting the data takes too much time, so it is difficult to transmit the data in real time.
Therefore, the image of the object is coded, and the motion of the image is estimated and then the amount of the information in the image is reduced for storing the information in a predetermined recording/writing medium. Thereafter, the information is transmitted to a predetermined destination in real time.
When coding an image of an object, a vector quantumization method or a discrete cosine transform (DCT) method is used.
Recently, a shape adaptive discrete cosine transform (SADCT) method has been effectively used in industry. This method is very effective for object-based compaction.
The above-mentioned shape adaptive discrete cosine transform method is directed to forming a grid with respect to an image frame, dividing an image of an object into a plurality of unit regions each having predetermined size and shape information, separating a unit region from the plurality of the unit regions in which an image of the object exists and then coding the unit region.
In addition, when the unit region contains an image to be coded, the effectiveness between a two-dimensional region DCT and a compaction become identical in the shape adaptive discrete cosine transform. When the unit region does not contain the image to be coded, the pixel, in which an image of an object exists, is processed with respect to the X-axis in a one-dimensional discrete cosine transform method, and a result of the above X-axis-based process is processed with respect to the Y-axis in a one-dimensional discrete cosine transform method. Thereafter, the final result value is obtained.
The shape adaptive discrete cosine transform method is further directed to reducing the number of unit regions in which an image of an object exists and performing the compaction after substantially filling the image of the object in the unit region, thus enhancing the compaction of a transform constant.
Therefore, when performing the shape adaptive discrete cosine transform process, the image of the object to be coded should preferably be substantially filled in each unit region, and then the number of unit regions in which an image of the object exists is effectively reduced.
The above-described shape adaptive discrete cosine transform process will now be described in more detail with reference to FIGS. 1A through 3F.
FIGS. 1A and 1B show grid patterns formed in one frame.
As shown therein, one frame is divided into a plurality of rows and columns which are consisted of a plurality of unit regions 21 having the same size and shape in cooperation with a P.times.Q number of X-axis grid and Y-axis grid 11 and 13 spaced apart from one another at a regular distance.
A unit region 21 may be formed in various shapes.
For example, the unit region 21 is formed in a regular square or a rectangular form by the X-axis and Y-axis grid 11 and 13. In addition, as shown in FIG. 2A. A unit region 21 may be formed as a horizontally lying triangle or a horizontally upside down triangle, and neighboring triangles form rectangular shapes bounded by the slant grids 15 and 17. As shown in FIG. 2B, a unit region 21 is formed by vertically lying triangles and neighboring triangles form rectangular shapes bounded by the slant grids 15 and 17.
In addition, as shown in FIG. 2C, the unit region 21 is formed as a 45.degree. rotated square by the slant grids 15 and 17, and as shown in FIGS. 2D and 2E, the unit region 21 is formed in a hexagonal shape by the slant grids 15 and 17. As shown in FIG. 2F, the unit region 21 is formed in an octagonal form having a 45.degree. rotated small square between the neighboring octagons. In this example, two different shaped unit regions 21 and concurrently used.
Any shape which spatially and evenly divides the image frame may be used for the unit region 21.
A square- or rectangular-shaped unit region 21 which is defined by an X-axis grid 11 and Y-axis grid 13 will now be explained.
As shown in FIG. 1B, the unit region 21 is formed of an M.times.N number of unit pixels 23 in the X-axis and Y-axis directions. For example, one unit region 21 is formed of an 8.times.8 number of unit pixels 23 or is formed of a 16.times.16 number of unit pixels 23.
In addition, a unit region 21 is defined as an M.times.N number of blocks in accordance with the number of unit pixels 23. As shown in FIG. 1B, the unit region 21 refers to an 8.times.8 number of blocks corresponding to unit pixels.
FIG. 3A shows an image (shown as the hatched portion) having predetermined shape information in a unit region 21 formed of an 8.times.8 number of unit pixels 23.
For the shape adaptive discrete cosine transform with respect to the image of an object, as shown in FIG. 3B, the image of the object is filled from the upper side margin portion of the unit region 21, and then the one-dimensional cosine transform is performed with respect to the Y-axis which is shown in the vertical direction.
The one-dimensional discrete cosine transform is performed as shown in FIG. 3D.
When the one-dimensional discrete cosine transform is completed with respect to the Y-axis, the image of the object is filled from the left side margin portion of the unit region 21, as shown in FIG. 3E, and then the one-dimensional discrete cosine transform is performed with respect to the X-axis which is shown in the horizontal direction.
When the one-dimensional discrete cosine transform is completed with respect to the X-axis, as shown in FIG. 3F, the shape adaptive discrete cosine transform with respect to the Y-axis and X-axis is completed.
Thereafter, a zig-zag scan is performed with respect to the final shape, as shown in FIG. 3F, which is obtained by the above-mentioned shape adaptive discrete cosine transform. For example, the zig-zag scan is performed diagonally from the leftmost side and the uppermost side to the rightmost side and the lowermost side.
However, the conventional shape adaptive discrete cosine transform is directed to performing the shape adaptive discrete cosine transform in accordance with the position in which the image of an object exists without moving the position of the grid.
Therefore, the bit rate per frame is high, and since the number of the unit regions in which the image of the object exists is numerous, there is a restriction on the ability to reduce an amount of compaction information which is obtained by coding the image of the object and the amount of motion information which is obtained by estimating the motion of the object.
In addition, when coding an object in the conventional discrete cosine transform method or the vector quantumization, since the compaction is performed without moving the position of the grid in accordance with the position in which the image of the object exists, the hit rate per frame is high as in the shape adaptive discrete cosine transform, and since the number of unit regions in which the image of the object exists is numerous, there is a restriction on the ability to reduce the amount of compaction information and the amount of motion information.
Meanwhile, when coding the image of a moving object among the images of an object having predetermined shape information, an object-based moving image coding method is generally used in the industry.
The above-mentioned object-based moving image coding method is directed to segmenting the image of the object in a background in which there is not a moving image and a changed region which is defined by the moving image of the object.
In addition, the moving object of the changed region is segmented into a motion compensable object and a motion compensable failed object through motion estimation.
Here, the motion compensable object refers to the moving object having a predetermined theory such as a horizontal movement, a rotational movement, a lineal movement, and the like in a state that the object in a three-dimensional space is converted into a two-dimensional image of the object. In addition, the motion compensable failed object refers to an object which is not adaptable with respect to the above-mentioned theory.
When transmitting and storing the image of the object, the motion compensable object process is directed to detecting motion information of the image of the object.
In addition, the image of the motion compensable failed object and the image of the exposed object are most effectively coded so as to reduce the amount of information, which is then transmitted and stored.
Since the amount of information with respect to the image of the motion compensable failed object is about 60-70% of the total amount of the information to be transmitted, many studies have been conducted, in the industry, so as to reduce the amount of information transmitted.
The motion estimation of the motion compensable object is directed to segmenting and estimating the moving portion of the moving image from a picture of the previous frame so as to minimize the amount of motion information.
However, since the variables with respect to the moving object are various, it is difficult to effectively extract, transmit, and store motion information in response to the immediate movement of the object.
Therefore, in the industry, it is urgently needed to transmit and store picture information of a high resolution having a small amount of information with respect to the motion compensable object in the motion estimation method.